Continuing the interview with Djihed Afifi, maintainer of the GNOME localization for the Arabic language, he talks about the Arabeyes Project, which the GNOME Arabic l10n team is part of, and the peculiarities of translating to the Arabic language.
What’s the story behind Arabeyes?
I wasn’t really around when Arabeyes started. From what I understand, prior to 2001 there were some indiviual and small organisational efforts for Arabic unix support, and Arabeyes was started in 2001. [It was created by] various people from around the whole world, mainly Arabic speaking but we do have a few non Arabic speaking volunteers. In general, [we aim] for Arabic support on Free and Open Source software. We try to report arabic related bugs, submit patches, translate projects and develop new Arabic related projects.
I think there is a wikipedia entry for Arabeyes that provides some more info. It was written by one of the old timers.
Some projects, like GNOME, KDE and to a lesser extent GNU, offer some infrastructure for their localization teams: statistics, version control, bug tracking etc. How much does Arabeyes use its own tools, and how much does it use the “target project” tools?
We use a mixture of everything that facilitates our work, both “in-house” and upstream 🙂
In general, we have our own infrastructure at Arabeyes with CVS and mailing lists and a wiki. The mailing lists serve the usual role of communication between developers/translators and sometimes upstream developers. We use our own CVS so people can work with ease with each other’s translations, fixing them, completing them, etc. We use the wiki to discuss technical dictionary terms and some of the more formal linguistic rules. Finally, there is a point of contact for every project that keeps in tabs with upstream, in general, by syncing translations between our CVS and the project’s repositories, echo’ing announcements and deadlines, filing bugs and keeping a watch on them, etc.
And where appropriate, we use tools upstream where they help us. For example, we use damned-lies extensively to assign files, lookup translation statistics and other tasks.
How does Arabeyes deal with translation standards and translation memory? What are the plans on that field?
For translation memory, in general people have use their own tools (poedit and kbabel are two examples). However, team leaders in general have a big translation memory and we use it to fuzzy translate as much as possible before giving the files to translators.
The technical dictionary is basically an English-Arabic dictionary for computing terms. At first, we started making it with .po files, but that created many problems with versioning and discussing the terms. So I had the idea of uploading the terms to our Wiki. I used some scripts to convert the .po files to Wiki xml input. The wiki, being open, allows people to edit as they see fit, discuss terms, suggest alternatives, etc. Then finally, there are some scripts that take the wiki pages and convert them back to .po files, as well as .pdf suitable for printing/reading. The experience was very rewarding to us.
Our main work is now centered around the technical dictionary, it is really the basis for other projects such as spell checking automatic translation (hopefully).
We also flirt with the idea of open translation for everyone using Pootle from time to time, and it looks like we may adopt it at some point in the future.
Besides translating software, what are the other Arabeyes projects?
Development ad documentation. We develop a few projects such as the ITL (islamic librari), Siraji (Arabic OCR in its infancy), and we try to patch some well known projects to offer support for Arabic (VIM, putty…).
What are your plans for Arabeyes and its GNOME translation team?
At the GNOME team, we are heavily translating the most common applications. For a long time, we haven’t touched big projects like the GIMP and the office applications. This is changing lately thanks to a team of highly motivated translators. I’d like to see most of the GNOME applications translated, then perhaps afterwards, we might wonder into the documentation world. Personally, I’m spending more time reporting and trying to fix Arabic and RTL bugs in GNOME applications. Some quirks still exist.
For Arabeyes, we are forever in need for contributors. We do think of lots of ideas, but we always hit the shortage of manpower wall. We’d like to see Arabic support addressed in all popular OSS applications. We’d also like to develop a free Arabic OCR application and an automatic translator. This is short term, but the long term list is a big one.
They are very good efforts, although the Arabic distribution sphere is very segmented and inefficient, and I have been trying for some time to draw people together to unify their efforts.
According to Wikipedia, the Arabic language has “27 sub-languages” and is spoken in numerous countries. How do the Arabic translation teams deal with that?
Dialects, in general, are very informal and are not used at all in education and computers. Think your typical street informal dialect, only the Arabic ones are a bit further from the formal language.
If you mean locales, we don’t deal with those as they require too much work for little gain, the main differences are the way the different countries name, months, currencies, etc.
Would the Arabic translation teams ever start a “meta project” if the language was closer to English, like Portuguese?
I am not sure to be honest 🙂
Arabic in general is a demanding language in terms of support in programs. Arabic is RTL, has a completely different characters set, needs shaping, joining and other things. But a unified team helps a lot for languages that do not have a lot of contributors. In general, more or less the same group of people translates most of the other projects, so the meta project brought people together and served as a point of contact and interest for everybody.
That’s it! Thanks to Djihed for the interview.
I should post once in a while other interviews about free software localization, both in Brazil and around the world. Stay tuned!